B2B, Data visualization, Product strategy, design system
Google X
AI-driven nationwide electrical grid planning tool that optimizes management, enhances performance forecasting, and visualizes data for decision-making
View More Details
Plant Daddy
Duration
Apr 2022, 3 Weeks
Tools
Arduino Moisture Sensor, LED Lights, LED Light Bar, Speaker, Teachable Machine
Role
Product Designer, Developer
Description
This is a 3-member group project for the Prototyping class. This was the first time we learned how to use Arduino to do some interactive prototyping. We designed and prototyped a smart plant pot that reminds users to water the plant when needed and keeps pets from damaging the plant.
I was responsible for the Arduino programming and training the teachable machine model.
Overview
Plant Daddy is a hardware prototype made for the plant's mom and dad.
This smart plant potted plant tells the current water intake of the plant, reminds users to water the plant when needed, detects any pet movement around the plant, and keeps pets from damaging the plants with voices.
We use moisture sensor to detect the wetness of the soil and train a machine learning model with Teachable Machine to determine if a pet is present and use a speaker to make a sound to make the pet leave.
Ideation
Because of millennials, plantfluencers, plant-based movement, and work from home culture forced by the pandemic, 7 in 10 millennials call themselves a plant parent and have killed seven houseplants on an average.
During our research, we found that many plant parents had another problem, that is, their pets would often knock over the plants or eat them.
From this, we have identified our target users.
Main feature:
• Tells you about the current water intake of your plants and how much is actually needed in real-time.
• Reminds you to water the plants when needed.
• Detects any pet movement around the plant.
• Keep your pet animals from destroying the plants through voice assistance.
Process
We use a moisture sensor to detect the wetness of the soil and train a machine learning model with Teachable Machine to determine if a pet is present and use a speaker to make a sound to make the pet leave.
We first tested the moisture sensor to make sure it read data properly and explored possible output methods to display humidity. Ultimately we decided to use a light strip to display the real-time moisture and the recommended moisture.
Our next step was to implement the other main feature, which is also the most challenging part, how to recognize pets and the background environment. We have tried many different methods and finally decided to try the Teachable Machine.
The teachable machine was new to us, and we first tried to learn how to train a model with it. In the image on the left, we need to input numerous pictures to tell the machine what a human face is and what a book is.
After this, we trained the machine to recognize cats, dogs, and other distracting items as background.
We mark the results of the 3 different machine learning recognitions as 3 different states and send the signals to the Arduino Adafruit Circuit Playground. The Arduino board will output different sounds and LED light changes based on this.
After completing the code, we created the final product prototype. We wanted to simulate the actual product and usage environment as much as possible, so we got a natural plant.
Excitingly, our product worked well and helped the plant parents keep the plant alive!
